A Virtual Course on Automation of Agriculture

8/6/2019 A Virtual Course on Automation of Agriculture

1/13

A Virtual Course on Automation of Agricultural SystemsFRANCISCO RODRI GUEZ, MANUEL BERENGUEL AND JOSE LUIS GUZMA NUniversidad de Almer a, A rea de Ingenier a de Sistemas y Automa tica, Ctra. Sacramento s/n, 04120,Almer a, Spain. Email: [email protected]

SEBASTIAN DORMIDOUniversidad Nacional de Educacio n a Distancia, Dpto. de Informa tica y Automa tica, C/. Juan del Rosal 16, 28040, Madrid, Spain.

New teaching methods and innovative techniques have recently appeared that aim to enhancestudents' motivation and improve their education. This paper describes the experience of developing a virtual course on modern automation of agricultural systems. It includes both classical teaching tools and novel teaching methods, taking advantage of new information and communicationtechnologies (ICT). The course is based on the WebCT platform and includes interactive tools

and both a virtual and a remote laboratory for greenhouse climate control and fertirrigationteaching/learning.

Keywords: ICT; WebCT; virtual course; agricultural systems; control; automation

INTRODUCTION

AGRICULTURE has been one of the less techni-cally developed sectors as a result of a number of problems, such as socio-economic limitations(such as division of the land into tiny farms, littlecapital investment and the low value of the end

products), the seasonal cycle of the crops andcheap low-skilled manual labour. Furthermore, itis influenced by the huge variety of crops andinstallations, as well as by adverse and variableorographical and meteorological conditions.Nevertheless, the development of utensils, toolsand sophisticated machines to help producefoods and raw materials has been a major preoc-cupation within the field of agricultural over thelast years. Nowadays, the agro-alimentary sector isthe object of special attention, and many initiativesare being developed to incorporate new technolo-gies, aimed at increasing production, and achievingdiversity, quality, and market presentation require-ments, as well as solving the problems of thedeficiencies in and scarcity of the manual labouravailable.

So, technological renovation of the agriculturalsector is required and control engineering can playa decisive role. It incorporates automatic controland robotics (AC&R) techniques at all levels of agricultural production: planting, production,harvest and post-harvest processes and transporta-tion. This fact is corroborated by major agricul-tural engineering associations that coordinatework groups in these subjects, such as the Inter-national Commission of Agricultural Engineering

(Section VII: Information Systems), the Interna-tional Society for Horticultural Science (Workinggroups in Sensors in Horticulture, GreenhouseEnvironment and Climate Control and ModellingPlant Growth), the European Society of Agricul-tural Engineers (Special Interest group in Auto-mation & Emerging Technologies and Information

Technologies) and The Society of Engineering inAgriculture, Food And Biological Systems (Infor-mation and Electrical Technologies Division).Moreover, a society devoted exclusively to theapplication of the new technologies in agriculturewas founded in 1996: The European Federationfor Information Technology in Agriculture, Foodand the Environment. On the other hand, the mostprestigious associations in automatic control androbotics consider the agricultural sector to be high-priority. As an example, the International Federa-tion of Automatic Control has a TechnicalCommittee called the Control in Agriculture andthe IEEE Robotics and Automation Association,which supports a Technical Committee on Agri-cultural Robotics .

Because of this, introductory courses of AC&Rare being included in modern agronomy studies.Future agricultural engineers will have to use,understand, and implement the new advances inautomation in the sector. In fact, a number of international organizations, such as the Interna-tional Commission of Agricultural Engineering(CIGR) [1], AFANet (EU Thematic Network)[2], FEANI (Federation Europeenne D'Associa-tions Nationales D'Ingenieurs) [3] and USAEE(University Studies of Agricultural Engineering

in Europe) of the EurAgEng (European Associa-tion of Agricultural Engineers) [4], have made* Accepted 4 August 2006.

1197

Int. J. Engng Ed. Vol. 22, No. 6, pp. 11971210, 2006 0949-149X/91 $3.00+0.00Printed in Great Britain. # 2006 TEMPUS Publications.


2/13

suggestions regarding the subjects that have to betaught in modern agricultural studies with the aimof globalizing agricultural engineering education.Some recommendations focus on the inclusion of classical engineering courses dealing with theanalysis and design of control systems, as well asmodern courses on systems identification, modern

control (optimal, predictive, and robust control),robotics, and image analysis.

This paper describes the development of avirtual course on modern automation of agricul-tural systems, called Automatic Control andRobotics in Agriculture, and includes both classi-cal teaching tools and novel education methods,taking advantage of new information and com-munication technologies (ICTs). The course hasbeen developed within the scope of an agriculturalengineering Ph.D. programme in the University of Almera (South-East of Spain) and taught inSpanish to students of a number of Latin Ameri-can countries (Venezuela, Mexico, Colombia, etc.),who had different a priori knowledge on thesubjects included in the course. The main featuresof this course are discussed, including the frame-work, objectives, the method of grading, etc. Theeight units involving the basis of AC&R andmachine vision techniques applied to differentagricultural fields are also described. Their aim isto motivate students with both a strong theoreticalbase and engineering skills, which is a majorchallenge in engineering education. The nextsection is devoted to comments on the new ICTsthat are used in this course, for virtual teaching,interactive tools, and virtual and remote labora-

tories. The last sections include feedback from thestudents and some concluding remarks.

COURSE DESCRIPTION

Framework Automatic control and Robotics in Agriculture

is a course included in the Ph.D. programmeAgroplasticulture, Agronics, and SustainableRural Development, offered by the University of Almera (Spain).

The province of Almer a has the largest concen-tration of greenhouses in the world (covering morethan 35 000 hectares). Because of this, agriculturalengineering studies are probably the most repre-sentative studies in this University, supported bywide educational and research experiences in thisfield. So, this Ph.D. program has been developedto satisfy the requirements formulated by someLatin American Universities and AgriculturalResearch Centres. Their interest is in providingtheir future doctors with a strong theoretical andpractical knowledge of agroplasticulture (croppingunder plastic greenhouses), on the use of the newtechnologies in agricultural production in tropicalregions, and on the sustainable development of

their rural communities.The programme offers seventeen courses lasting

20 to 30 hours each, covering the followingaspects:. automatic control, electronics, and robotics

applied to agricultural processes;. agrarian economy, markets investigation, and

rural sustainable development; and. management of installations and production

engineering in greenhouses.

Spanish Ph.D. courses are divided into twoyears. During the first year the students mustselect and attend a minimum of seven coursesand during the second year they must produce asupervised research work that will be the basis of their Ph.D. thesis.

As the Latin American students cannot physi-cally attend the courses due to the high costsinvolved, a virtual course has been developedthat includes remote and virtual laboratories,interactive tools, and classical distance learningelements. In this way, the students have to travelto Almera only once, to present the research workproduced during the course.

ObjectivesBy following this course the students will be able

to do the following.Acquire a solid base on AC&R technologies to

help them:. analyse controllable systems, recognize their

elemental parts and the techniques used intheir design;

. understand the evolution that the automationsector is undergoing;

. use basic computer-aided engineering tools formathematical calculations in engineering, simu-lation and programming.

Study the agricultural tasks where AC&R tech-niques can be implemented using elements such asProgrammable Logic Controllers (PLCs), compu-ters, and robots.

Use the learned concepts and techniques toanalyse the commercial systems currently used inagriculture, compare them by studying theiradvantages and disadvantages, and select them tofulfil the required features.

ContentsIn order to select the contents of this course, an

ideal academic two-month period has been consid-ered, composed of 25 hours for theoretical lessonsand 15 exercises, as well as laboratory hoursdistributed over 15 weeks. This schedule is fairlyflexible; in fact, a dilemma appeared when prepar-ing it: we had to decide whether the basic subjectsof AC&R should come before or after thosedevoted to applications in agriculture, or if thecourse should be divided into two modules (onedevoted to automatic control and the other torobotics). Based on the experiences of traditional

lecturers during the academic courses 1995/96 to2005/2006, where both approaches had been used,

Francisco Rodr guez et al.1198


3/13

it has been observed that the students preferred thesecond scheme. This is because AC&R disciplinesare not directly related to their acquired knowledgeof agronomy. So, if both areas are mixed duringthe programme that shows the applications of theAC&R techniques to agricultural processes andtheir fundamentals, the course is easier for the

students to understand.The professional and research experiences of

the authors of this course are oriented towardsgreenhouse crop production (agriculture underplastic cover). This agricultural productionmodel is being exported to many countries,mainly in the Mediterranean area and LatinAmerica. Because of this, a large part of thecourse is focused on this crop production modeland its levels (plantation, production, harvesting,post-harvesting and transport). The applicationof AC&R to the traditional agricultural sector isalso covered in the course: the harvesting of fruitsfrom trees or autonomous tractors, with severalrelevant references.

Based on these considerations, the contents of

this course are divided into two modules eachcomposed of four units, as Table 1 shows.

Study guideThe following information and material are

given to the students in each of the eight learningunits of the course.

. Objectives . Summary of the basic objectives of the unit and the real related problems that thestudents should be able to solve.

. Activity planning . This includes a description of the unit sections and the timing of each one.

. Documents of the unit :. Contents . The unit contents are supplied in

html format files, including hyperlinks toother related parts of the documents andinteresting URL addresses. A table of con-tents placed at the beginning of the unitprovides direct access to the contents in an

easy way.. Exercises . A set of problems that the studentsmust solve are proposed. All of them deal

Table 1. Contents and timing of the course

Module 1 Automatic control applied to agricultural systems

Unit 1. Modelling and simulation of agricultural systemsThe design of a control system requires the knowledge of its dynamical behaviour. It is thus necessary to design and implement a modelof the system, consisting of a mathematical description of the relations between the controlled variables and the input (disturbance andcontrol) variables. The use of a dynamic model involves the key questions in the design of a control system: coupling of the variablesforming the control loops, definition of the limits in the control action, evaluation of the control difficulty (sign and sensitivity of thecontrol actions, speed of response and shape of the controlled variable), and the sensitivity to changes in the process as a consequence of setpoint shifts and disturbances.

Unit 2. Automatic control of dynamic systemsThe main objective of a control system is the maintenance of a variable into a desired interval with a determined behaviour. There areseveral processes in the agricultural sector controlled by automatic systems, like the pH of the irrigation solution. Thus, the generalprocedure to design the control system is described in this unit based on the classical PID algorithms.

Unit 3. Sequential control of processesThere are agricultural processes in which it is not sufficient to control the dynamics of the variables, but also the sequential transitionsfrom a state of the system to another. This unit is thus devoted to studying the basic issues involved in the sequential control of agricultural systems using PLCs (Programmable Logic Controllers).

Unit 4. Automatic control of agricultural systemsThe crop growth is determined by the climatic variables of the environment, the irrigation and the supply of fertilizers. Greenhouseclimate and fertirrigation control systems are described in this unit, showing the sensors, actuators, and the different control algorithmsused in commercial and research equipment.

Module 2. Robotics applied to agriculture

Unit 5. Manipulation roboticsRobot manipulators are indispensable tools for performing repetitive and dangerous tasks, substituting for human labour. Somestudies have identified several agricultural tasks that can be carried out by robot manipulators, like the harvesting process. This unit isdedicated to the fundamentals of robot manipulators (elements, control and programming) and has been included in order tounderstand this technology and its possible applications in agriculture.

Unit 6. Mobile roboticsA feature that has contributed to the development of mankind has been its ability to move by itself or by making use of other elementsto increase its speed (animals, wheels, motors, etc.). Vehicles allow the displacement of great masses at relatively high speeds, but theyrequire a human worker for the arduous task of driving. Autonomous mobile robots can avoid this problem, as they are characterizedby their capacity to move themselves in known (and even unknown) environments without a human decision. Mobile robotics is beingused in the field of agriculture for specific tasks (e.g. spraying robots) or for augmenting the autonomy of agricultural machines (e.g.tractors or combine harvesters). These reasons justify their inclusion of this unit.

Unit 7. Machine vision applied to agricultureMachine vision is widely used in the agricultural sector nowadays in applications such as fruit sorting in post-harvesting processes,quality control of the seeds germination in nurseries, or fruit location in harvesting processes. This unit shows the elements and the sixprocessing levels in a machine vision system (acquisition, pre-processing, segmentation, description, identification, and interpretation)together with more usual agricultural applications.

Unit 8. Agricultural robotsAs mentioned before, agriculture is a sector where robotics (manipulation and mobile) is being widely used. This unit shows bothcommercial and research robotic developments in order to increase the autonomous level of agricultural machines, production harvestin trees and plants, or processes like transplanting or cuttingsticking.

Virtual Course on Automation of Agricultural Systems 1199


4/13

with an AC&R application in the agriculturalfield. The exercises include practical applica-tions and situations that are very familiar tothe students, in order to help them assimilat-ing the studied concepts and techniques thatare unknown for them at the beginning of thecourse. These exercises must be sent to the

teachers via email, constituting the base of thegrade.

. Additional readings . A set of papers (Journals,Conferences, and Congresses) and spreadingslides presentations developed by the teachersare supplied to the students, includingresearch and practical experiences related tothe contents of the unit.

. References . This box comprises the basic andcomplementary bibliography used to elabo-rate on the documentation of the unit.

. Interactive and virtual tools. Many differentinteractive tools and virtual laboratories havebeen developed by the authors and some of themare used in this course to help the students tounderstand the evolved concepts and techniques.These tools will be commented on in the nextsection.

The students have also to do two practices usingremote laboratories located at the automaticcontrol, robotics, and machine vision laboratoryof the University of Almer a. The first deals withthe design and implementation of a greenhousetemperature control algorithm using natural venti-lation and heating systems, while the second isdevoted to the design and implementation of analgorithm to manage a robotic cell to pick andplace pots of ornamentals flowers in a greenhouse.

The main method of communication betweenthe teachers and the students is by email, throughwhich all queries and questions relating to thetheory, problems, exercises, and practices can behandled. Furthermore, two debate chat forums areprogrammed during which all the comments (rela-tive to the concepts treated in each one of the twomodules) are discussed. In this way, a traditionalclassroom can be emulated, because all the courseparticipants (teachers and students) are connectedin real time `talking' about the various applicationsof AC&R in several countries with different agri-

cultural production models and sharing experi-ences. Participation in these chats is interesting asit enables questions to arise that could otherwise beignored.

Course gradeThe course grade is achieved based on the three

following activities performed by the students.

1. Exercises . The students must work through theset of exercises proposed in each one of thelearning units and send the solution to theteachers via email for their qualification.These are evaluated between 0 and 10 points,

and all the students must attain a minimum of 5points in each of the units. The evaluated and

corrected exercises are fed back to the students,accompanied by comments.

2. Remote laboratory practices . As has beenpointed out above, two practices are proposedat the end of each module, related to AC&Rapplications in agriculture. Both are evaluatedbetween 5 and 10 points, and serve to modulate

the qualification obtained in the exercises.3. Discussion forums . At the end of each module, a

discussion forum is proposed to find out theopinion and comments of the students andteachers about the course. Participation inthese activities is also considered to fulfil thegrade of the course.

NEW TECHNOLOGIES APPLIEDIN THE COURSE

In the last decades our society has been seenhuge advances in ICT, giving rise to innumerablechanges in enterprise, cultural, social, and educa-tional fields. These advances have promoted thedevelopment of new techniques and methods,opening a wide range of possibilities not only inthe industrial area (remote control, remotemanagement, flexible timetable, etc.) , but also inthe field of education (distance education, inter-active tools, remote and virtual laboratories, etc.),the rapid evolution of computers, the Internet,distributed computing facilitates and the develop-ment of feasible and cost-effective solutions.

The main advantages of ICT are related toteleaccess, teleoperation, and telecontrol. Until

recently, the exchange of information was doneusing local networks for several reasons: informa-tion safety, narrow bandwidths, limited tools forthe exchange of information, etc. Thanks to theadvances in Internet technologies, a new methodfor accessing the information has appeared, theteleaccess, providing a way to access the informa-tion safely from any part of the world withouttemporal constraints. Both in the industrial andresearch/educational fields, high costs are oftenrelated to the restricted use of (usually expensive)systems with usage-time constraints, or to thedisplacements required to control them. Withteleoperation (extension of sensorial capabilitiesand human skills to a remote place) and telecontrol(the specific part of teleoperation whose goal is tosend commands to the actuators) technologies, it ispossible to control systems remotely through theInternet, thus helping to reduce displacement costsand allowing one to extend the use of time-limitedresources or equipment.

New teaching methods and innovative techni-ques have recently appeared, aimed at enhancingstudents' motivation and improving their educa-tion: multimedia tools, hypertext systems, interac-tive systems, information exchange betweenteacher and student through the Internet, access

to informationfrom any part of the world withouttemporal constraints, etc. All these methods are



5/13


6/13

Interactive tools for teaching AC&RInteractive tools are one of the most remarkable

advances that have arisen in the educational field,thanks to unprecedented challenges in the newtechnologies. Interactive tools are powerfulelements that help the students to enhance theirskill, motivation, and ability to understand and

solve control problems, attempting to `demystify'abstract mathematical concepts through visual-ization by using specially chosen examples [7].The AC&R theory covers many topics and agood engineer must handle the basic concepts,techniques and ideas, and must be able to use allof them in real problems. The need for providingstudents with both a strong theoretical base andengineering ability is a major challenge in AC&Reducation [8].

As commented in [7], the continuous increase incomputer processing power allows the use of interactive tools to merge the synthesis and analy-sis phases of classical engineering iterative designtechniques into one. In this way, the modificationof any configuration, parameter, or element in thesystem produces an immediate effect in the rest,and thus the design procedure becomes reallydynamic and the compromises that can beachieved are identified more easily.

Many tools for control education have beendeveloped over the years, incorporating theconcepts of dynamic pictures and virtual interac-tive systems [9]. Nowadays a new generation of software packages has appeared, based on objectsthat admit a direct graphic manipulation and areautomatically updated, so that the relationships

between them are continuously maintained. Ictoolsand CCSdemo [8, 9], developed in the Departmentof Automatic Control at Lund Institute of Tech-nology or SysQuake in the Institut d'Automa tiqueof the Federal Polytechnic School of Lausanne [10]are good examples of this new educational philo-sophy of teaching.

The authors of this course have experience inthis field (e.g. [7, 11, 1214]) and reduced versionsof three interactive tools have been used toenhance students' knowledge and skills in model-ling, control and robotics, evolving only the

concepts that an agricultural engineer requires.These tools are sent to the students via email andcan be obtained by contacting the authors. Oneconsideration that must be kept in mind is that thetool's main feature, interactivity, cannot be easilyillustrated in a written text. Nevertheless, some of the advantages of the application are shown below.

Systems modelling and analysis tool The tool has been developed using Sysquake [10]

and was created using the huge library of inter-active components developed by ProfessorDormido. This tool helps students to perform thestudy, analysis, and simulation of dynamic systemsinteractively. The application is available in twoversions, one for first order systems (Fig. 1(a)) andother for second-order systems (Fig. 1(b) ). Thereason for paying attention to first- and second-order dynamical systems is twofold. First, thesekinds of systems described by ordinary first- andsecond-order differential equations (which have adirect interpretation in the Laplace domain usingblock diagrams based on the transfer functionconcept) are described in all the classical controltextbooks and can be found elsewhere (e.g. [15,16]). Secondly, many agricultural systems, whenlinearized around an operating point, can bedescribed as first-order linear dynamical systems(e.g. dependence of greenhouse temperature withvents opening or heating, water level in tanks, etc.)or second-order ones (e.g. valves, level of tanks inseries, etc.). Descriptions of these can also be foundin agronomical textbooks (e.g. [17, 18, 19]). Usingthe tool it is possible to modify the different

characteristic parameters of the systems using theknobs or hovering directly over the graphs, and toobserve the dynamical responses and the poleplacement of the systems (which determine thebehaviour of the system) immediately. Somecomments regarding nonlinear modelling areincluded in the course and related references.

Analysis and design of PID (proportional,integral, and derivate) controller's tool

Once the students have the necessary knowledgeof analysis and modelling dynamical systems and

Fig. 1. Interactive tools for modelling and analysis of systems: (a) first-order systems; (b) second-order systems.



7/13

design of classical controllers, they can use anotherinteractive tool (an extension of the previous one)to see the effects of the PID algorithm whencontrolling agricultural processes. The action of this controller is based on the error, error integral,and error derivative between the desired trajectoryand the real process output. Several applicationexamples, such as pH control of the solution of fertilizers, greenhouse temperature control, etc.,are shown. A description of this kind of controlalgorithm can be found in any control book andeven in several agricultural textbooks (e.g. [15, 16,18]). The tool allows one to interactively re-tunethe controller, compare the open-loop and closed-loop responses, study the response to disturbances,and perform the typical analysis of the controlsystem design. The tool is available in two

versions: for first order systems (Fig. 2(a) ) andfor second order systems (Fig. 2(b)).

Virtual lab for teaching greenhouse climate control The main aim of the tool is to facilitate the study

and understanding of the main issues involved inmodelling and climatic control of greenhouses.The tool consists of a graphic screen connectedto the algorithms and a model implemented inMatlab/Simulink [20]. This model is used by thestudents to understand the process dynamics inorder to obtain the model parameters. In this way,they can design the controllers and test them in thevirtual lab. The tool is shown in Fig. 3, where thelaboratory screen has been divided into three parts[21]: greenhouse synoptic, a set of graphs to showthe different variables, and several buttons andsliders to allow the modification of the controlparameters.

The greenhouse synoptic is shown in the leftupper part of the main screen. In this area it ispossible to monitor the crop growth and the effectof the control signals through the actuators' move-ments: natural ventilation, shade screen, and heat-ing based on a boiler. The lateral and roof ventilation positions are changed taking intoaccount the value of the associated control

signal. The actual value of the vents opening isshown as a percentage below the roof ventilations

and next to the right side ventilation. The shadescreen is located between the greenhouse and itscover; its length can be modified based on theassociated control signal. The third actuator isthe heating. This system has been represented bya boiler, a pump and several pipes. A colour codeis used: pump and boiler off/on (blue/green); pipetemperature (grey to red, based on its value). Thenumerical value of the pipes temperature is shownnext to the boiler, as can be seen in Fig. 3, where anexample of heating is shown. The crop growth isconsidered as a disturbance for the system, and it ispossible to modify this parameter in the graphicalscreen. Several plants are shown in the synopticwhere their size is changed, based on the value of this disturbance.

Regarding the other parts of the tool, several

graphs with different variables are shown on theright of Fig. 3. From left to right and up to down,the graphs contain the following information: (1)variables related to the temperature evolution:inside temperature, outside temperature, and dayand night temperature setpoints; (2) informationabout the control signals: vents opening, shadescreen opening, temperature of the heating pipes,and integral signal of the heating controller toexplain anti-windup schemes; (3) informationabout the radiation: outside PAR radiation,inside PAR radiation, radiation setpoint, and thedead-zone for shade screen controller; (4) windspeed and (5) the relative humidity. The evolutionof the outside temperature and wind speed areimportant for control purposes, as the propor-tional gain of the gain scheduling controller forthe temperature control with ventilation shown inthe last graph depends on these variables.

The last part of the tool corresponds to thecontrol parameters. This window is located belowthe greenhouse synoptic as seen in Fig. 3. In thefirst column there are three buttons, three editfields, and one checkbox. The buttons allow initi-alizing, pausing, and reinitializing the simulation;in the edit fields it is possible to modify thetemperature and radiation setpoints, while the

checkbox is used to start the simulation duringthe day or the night. The next two columns contain

Fig. 2. Analysis and design of PID controllers: (a) first-order systems; (b) second-order systems.



8/13

the control parameters related to temperature

control with ventilation using a gain schedulingcontroller, to account for the influence of the windspeed and the outside temperature in the calcula-tion of the controller parameters. The other para-meters of these columns allow one to set themaximum opening for the ventilation when thewind exceeds the maximum speed, and also tomodify the parameters associated to the setpointmodification based on the humidity. With respectto the last column, the first three parameters areused to modify the parameters of a PI+anti-windup temperature controller using heating. Thelast two parameters allow modifying the effect of the crop growth based on Leaf Area Index (LAI)and the dead zone of the on/off controller for theradiation control with shade screen, respectively.The control algorithms are described in [19] and[20].

The environment Easy Java Simulations (EJS) isa software tool designed for the creation of discretecomputer simulations [22]. It was used to developthe tool, because it is easy to use, the developmentlanguage is available in many platforms (Java), itcan be connected to Matlab/Simulink, and verygood results have been obtained in previous works.

Manipulation robotics tool

The advances experimented with in the inclusionof interactive tools within the robotics education

framework have not been as significant as those in

the automatic control field. However, many localsimulation tools and remote and virtual labs havebeen used, as is shown in the next section.

A tool has been developed (in Spanish) aimed atdefining and programming a robotized cell inter-actively (see Fig. 4) [23]. This tool is used to learnthe basic concepts of manipulation robotics thatcan be encountered in basic textbooks (actuators,sensors, morphology, kinematics modelling andcontrol, robot programming, etc., e.g. [24]) usingtypical cell robots and programming languages.The following operations can be performed withthe tool.

. The configuration of the simulation screen . Theuser can define the number of cell views. Thereare several views configured by default: centralview, left view, etc. It is also possible to changethe shading view: solid, meshed, or softened.

. The definition of the simulation environment . Theapplication allows defining personalized envir-onments with several objects and devices.

. The definition of the robots positions . It is pos-sible to move the robot interactively with knobs,dragging over the robot with the mouse, orintroducing the desired positions from the key-board. The user can record the different posi-

tions that have been reached by the robot at anytime.

Fig. 3. Main screen of the virtual lab for teaching greenhouse climate control.



9/13

. The running programs . Once the cell elementsand the robot positions have been defined, it ispossible to run a program in the new virtual cell.The program can be run step by step or in acomplete cycle.

. The visualization of information . This screenshows the information of the different elements:final position of the robot, values of each articu-lation coordinates, input and output states of the controller, etc.

The main features of this tool are based on arobotized cell that is installed in the University of Almera, which is described in the next section.

Teaching AC&R using WWW Sometimes it is necessary to incorporate

contents with a strong experimental componentand let the students apply their acquired know-ledge to real systems. Traditionally, local simula-tion tools or local laboratories have been used forthis purpose, but with space and time constraints.Nowadays, and thanks to advances in ICT, espe-cially in Internet technologies, the laboratoryenvironment can be transferred to distance educa-tion. Based on these premises, two new conceptshave appeared within the distance educationframework: virtual laboratories and remotelaboratories. The first of these are a new kind of simulation tools that is much more powerful thanthe traditional ones, allowing the simultaneous useof remote simulation modules by the students. Onthe other hand, remote laboratories permit thestudents to perform the main laboratory activitiesremotely, without requiring their presence at theplace where the hardware is placed, in such a waythat they can control and monitor physical devices24 hours a day at any time and anywhere, inter-acting with the teacher without having to travel tothe university. Remote labs help the students toput into practice the concepts that they havelearned by remotely accessing real systems.

In order to provide this course with the strongpractical component that it requires, two remote

labs have been used. One important aspect of aremote laboratory is for students to feel that theyare controlling a real system. One way to do this isby using visual feedback. Because of this, a videoserver (AXIS 2400) and two charge coupled device(CCD) cameras have been used to view the remotelabs that are presented in the next two subsections.

Web-based remote control laboratory using a greenhouse scale model

Taking into account the contents of the course,it was considered necessary to have the opportu-nity to perform quasi-real tests using a greenhousewithout temporal and spatial constraints. Keeping

this in mind, a greenhouse scale model (Fig. 5) wasdeveloped under the framework of the DAMOCIAproject (Computer-Aided Design for the Construc-tion of Automated Greenhouses) financed byESPRIT Information Technologies Programmeof the European Union (http://www.cordis.lu/esprit/home.html ) (Special Action P7510) [25].

Several sensors have been installed to measurethe main variables to be controlled: inside airtemperature and humidity and inside photosynth-esis active radiation ( PAR ). An external meteor-ological station has also been placed to measureoutside temperature, global and PAR radiation,humidity, wind speed and wind direction. Thesolar radiation variations are simulated using a500 W focus placed on the greenhouse scale model.Wind speed is also simulated using a direct current(DC) ventilator controlled via the serial port.There also exists the possibility of reading datafrom files obtained from experiences in real green-houses. From the control point of view, the scalemodel is provided with a natural ventilation system(with two DC motors for roof and lateral ventsaperture), a forced ventilation system (with a DCventilator), a simulated irrigation and fertilizationsystem (with leds), a simulated distributed pipesheating system (with leds), a heating system using

resistances, and a shade screen (with one DCmotor). The motors are activated using a set of

Fig. 4. Interactive tool for manipulation robotics: (a) main screen; (b) definition of objects and positions.



10/13

relays controlled by the serial port. Two controlmodules for serial ports (based on RS232 protocol)have also been used for acquiring the data.

A tool has been developed that incorporates theacquisition and control interfaces to allow thestudents to perform tests remotely [26]. A Websitefor the remote laboratory (http://aer.ual.es/

maqueta/) has been created, whose main page isshown in Fig. 5. The remote lab management toolwas developed using LabVIEW [27]. An exampleof the application is shown at the left bottom of Fig. 5.

The greenhouse scale model is used in thelaboratory sessions for implementing basic controltasks using the main actuators installed in green-house climate control systems and the fertilisersand irrigation elements. It allows the use of differ-ent types of controllers, because the greenhouse,the correcting equipment, and the power supplyconstitute an independent system. On the otherhand, different types of control algorithms can beimplemented in order to study the variableresponse of the actuators.

Using this remote lab the students can test themodelling and control results obtained analyticallyand with simulations (using the interactive toolpresented in the previous section). The algorithmcan be developed in Matlab [28] and at the end of the tests, a Webpage with graphical and text resultsis automatically created. An example of green-house temperature control using the heatingsystem is shown at the right bottom of Fig. 5,where the temperature tracks the setpoint trajec-tories.

Remote programming of robot manipulatorsAs mentioned in the previous section, it was

necessary to develop a remote robot programminglaboratory to test on a real robot cell the programsused in the simulation tests previously performedusing the interactive tool (see Fig. 4).

The robotic cell is shown in Fig. 6(a); it

comprises the following elements.. Mechanical arm . Robotic manipulator Scorbot

ER V-Plus of Eshed Robotec Ltd with fivedegrees of freedom.

. Multitasking controller in real time, whichallows several programs to run simultaneously.

. Teach pendant, which allows the control of the joints, the definition of the positions using theauto-learning method, and the execution of programs that are in the controller.

. A conveyor with an alternating current (AC)engine, which is controlled using a speed regu-lator connected to the robot controller.

. Photoelectric sensors to detect the presence of anobject on the conveyor.

. Vision system composed by a CCD colourcamera and a capture video card with framegrabber.

. Computer connected to the controller via RS-232 to manage the cell.

In order to access remotely to the system and totest the algorithms, a system developed in Java,Javascript and C++ Builder has been used [29] (seeFig. 6(b)). The students can write an ACL(Advanced Control Language) program via Web,can compile it and check the errors remotely. Once

Fig. 5. Greenhouse scale model, Website, and practice results.



11/13

the program is error-free, it can be tested in the realcell.

At present, there are many trends in greenhousecrop production of ornamental plants to optimizethe soil surface, so that the installation of conveyors to move pots to a work-station wherethe grower can carry out the basic operations onthe plants, such as planting or grafting, is required.In order to load and unload the pots, robotmanipulators are used. In the proposed exercise,the students have to design and implement aprogram that simulates this kind of work. Theyhave to detect the presence of a pot being carriedon a conveyor, and pick and place it at a deter-mined location, based on the input states of thesystem.

FEEDBACK FROM THE STUDENTS

To obtain feedback from the students, theteachers opened forums by asking a variety of questions regarding the schedule of the course,contents, etc. The general opinion of the students(11 answers were obtained) is that the quality andmaterial of the course are quite adequate, fulfillingwhat they expected. The use of interactive toolsand remote labs was surprisingly positive for them,as they had the opportunity of putting into prac-tice the knowledge they had gained. As weexpected, the main difficulty in following thecourse was in those units where basic concepts of modelling and control were introduced, as a math-ematical basis on differential equations andLaplace transforms was required (not all thestudents had the same previous knowledge thatthe students on these issues). Obviously, thoseunits devoted to AC&R applications in agriculturewere well accepted and understood by the students.The students also pointed out that the courseshould include more forums and even chatsessions, but from our view this is a quite time-consuming task and sometimes what the students

want from the teacher is to obtain the answer toexercises that they should obtain by themselves.

On the other hand, each of the students lookedfor different agricultural systems in their respectivecountries where the techniques learned during thecourse could be applied. In fact, some of thestudents intend to incorporate these aspects intotheir Ph.D. theses.

CONCLUSIONS

This paper describes a virtual online course onautomation and robotics applied to agriculturalsystems, and more specifically, to greenhouse cropproduction. It has been developed using ICTtechnologies for teaching all the related concepts

and techniques, because they facilitate learningand access to remote laboratories by the studentsfrom different geographical areas.

The theoretical part of this course wasdeveloped using hypertext systems, and it wascomplemented with several exercises. As pointedout in [7] and based on this experience, a generalrecommendation for automatic control androbotics engineering education is that students bemade aware of problems and potential solutionsarising from the increasing complexity of ouragricultural technological systems. Laboratoryexperience (real or interactive) is extremely impor-tant as a part of control learning. The goals of control education that might be kept in mindregardless of the specific choice of material or thestructure of the course are the following:. to provide the basis for lifetime learning so that

new control problems can be dealt with;. to establish and maintain high standards of

excellence for the experience of learning thefoundations of control.

From the authors' point of view, the teacher musttransmit his or her experiences and ideas to thestudents but not all; some of these ideas have tobe found out by the students themselves. The

hypertext systems, interactive tools, and remotelabs are very powerful elements, helping the

Fig. 6. Robotics remote lab



12/13

students to enhance their skill, motivation, andability to understand and solve engineeringproblems.

Acknowledgements This work was supported by the SpanishMinistry of Science and Technology (CICYT) under grantsDPI2001-2380-C02-02, DPI-2002-04375-C03-03 and DPI2001-1012.

REFERENCES

1. G. Pellizi, and P. Febo, The CIGR project for the harmonization of agricultural engineeringUniversity Curricula, Journal of Agricultural Education and Extension , 1(3), 1994, p. 6.

2. D. Briassoulis, H. Papadiamandopulou and B. S. Bennedsen, Towards a European Standard forAgricultural Engineering Curricula, AFANet, Denmark, (2001) p. 54.

3. FEANI, Guide to the FEANI register. EurIng, Par s, (2000) p. 15.4. USAEE, The Newsletter of the USAEE Thematic Network. USAEE, N0 4, p. 12 (2004)5. T. D. Urdan and C. C. Weggen, Corporate e-learning: exploring a new frontier, http://www.e-

learning.nl/publicaties/marktonderzoek/New_Frontier.pdf (2000) p. 95.6. WebCT Inc, Getting started Guide: WebCT Campus Edition. Ed. WebCT Inc., (2003) p. 94.7. S. Dormido, Control learning: present and future, Annual Reviews in Control , Wiley, 28(1), 2004,

pp. 115136.8. M. Johansson, M. Ga fvaert and K. J. A strom, Interactive tools for education in automatic control,

IEEE Control Systems Magazine , 18(3), 1998,pp. 3340.9. B. Wittenmark, H. Ha glund and M. Johansson, Dynamic pictures and interactive learning, IEEE

Control Systems Magazine, 18(3), 1998, pp. 2632.10. Y. Piguet, SysQuake: User Manual , Calerga (2000).

11. S. Dormido, M. Berenguel, S. Dormido-Canto and F. Rodr guez, Interactive learning of constrained generalized predictive control, 6th IFAC Symp. on Advances in Control Education ,Oulu, Finland, (2003) pp. 175180.

12. J. L. Guzma n, M. Berenguel and S. Dormido, Interactive teaching of constrained generalizedpredictive control, IEEE Control Systems Magazine , 25 (2), 2005, pp. 5265.

13. S. Dormido, The role of interactivity in control learning. Plenary Lecture, in Proc. 6th IFAC Symp.on Advances in Control Education , Oulu, Finland, (2003) pp. 112.

14. S. Dormido, F. Gordillo, S. Dormido-Canto and J. Aracil, An interactive tool for introductorynonlinear control systems education, 15th Triennial World Congress of IFAC, Barcelona, Spain,(2002) pp. 3338.

15. R. C. Dorf and R. H. Bishop, Modern Control Systems , 9th edn, Prentice Hall, (2000) p. 831.16. K. J. Astrom and B. Wittenmark, Computer Controlled Systems . Prentice-Hall, (1990) p. 488.17. J. Bontsema, et. al ., Greenhouse Climate Control. An integrated approach, Wageningen Press,

p. 276 (1995)18. Y. Hashimoto, G. P. A. Bot et. al ., The computerized greenhouses, Automatic Control Application

in Plant Production , Academic Press, (1993) p. 340.19. P. G. H. Kamp and G. J. Timmerman, Computerized Environmental Control in Greenhouses , IPC

Plant, (1996) p. 273.20. F. Rodr guez, Modelling and hierarchical control of greenhouse crop production (in Spanish:

Modelado y control jera rquico del crecimiento de cultivos en invernadero), Ph.D. Thesis.University of Almer a, Spain, (2002) p. 366.

21. J. L. Guzma n, F. Rodr guez, M. Berenguel and S. Dormido, Virtual lab for teaching greenhouseclimatic control, 16th World Congress of IFAC , Prague, Czech Republic, (2005).

22. F. Esquembre, Easy Java Simulations 3.1 .; (2002) http://fem.um.es/Ejs/23. J. F. Sarabia, F. Rodr guez, L. Iribarne and M. Berenguel. Simulation tool of a robotized cell, (in

Spanish: Herramienta de simulacio n de una celula robotizada), XXII Jornadas de Automa tica ,Barcelona, (2001) p. 7.

24. K. S. Fu, R. C. Gonzalez, and C. S. G. Lee, Robotics, Control, Sensing, Vision, and Intelligence,New York: McGraw Hill (1987)

25. F. Rodr guez, A. Corral, M. Berenguel and F. Bienvenido, Application scale model for teachingautomatic control of the greenhouse climate and irrigation tasks, 2nd European Conference forInformation Technology in Agriculture EFITA99 , Bonn, Germany, (1999) pp. 795805.

26. J. L. Guzma n, M. Berenguel, F. Rodr guez and S. Dormido. Web-based remote control laboratoryusing a greenhouse scale model, Computer Applications in Engineering Education , Wiley, 13 (2),2005 pp. 111124.

27. LabVIEW Distance Learning Solution Guide. National Instruments, http://ni.com/28. The MathWorks Inc., Using Matlab. The Language of Technical Computing, USA. (2002) p. 1188.29. J. L. Guzma n, J. F. Sarabia, F. Rodr guez, J. C. Moreno and M. Berenguel. Remote programming

environment of manipulator robots using ACL and with visual feedback. (in Spanish: Entornopara programacio n remota de robots manipuladores en ACL con realimentacio n visual), EIWISA 2002, Alicante, Spain, (2002) pp. 5763.

Francisco Rodr guez is an associate professor of systems engineering and automatic controlat the University of Almer a (Spain). He obtained his telecommunication engineeringdegree from Madrid Polytechnics University (Spain) and his Ph.D. degree from theUniversity of Almer a, in 2002. Currently he is a researcher and member of the AutomaticControl, Electronics, and Robotics group of the University of Almer a. His scientific

interests are focused on the application of modelling, automatic control, and roboticstechniques to agricultural systems and education.



13/13

Manuel Berenguel is an associate professor of systems engineering and automatic control.He is responsible for the Automatic Control, Electronics, and Robotics Group at theUniversity of Almer a, Spain. He obtained an industrial engineering degree and a Ph.D.from the University of Seville, Spain, where he received an award for the best engineeringthesis of the year. He was also a researcher and associate professor at the University of Seville for six years. His research interests are in the fields of predictive, adaptive, androbust control with applications to solar energy systems, agriculture, and biotechnology.

JoseLus Guzman obtained a computer science engineering degree in 2002 and a Ph.D. in2006 both from the University of Almer a, Spain, where he is a researcher of the Automaticcontrol, Electronics, and Robotics group. Currently his scientific interests are focus on thefield of control education and model predictive control techniques, with applications tointeractive tools, virtual and remote labs and agricultural processes.

Sebastian Dormido holds a degree in physics from the Complutense University in Madrid,Spain (1968) and a Ph.D. from University of The Basque Country, Spain (1971). In 1981, hewas appointed professor of control engineering at the Universidad Nacional de Educacio na Distancia. His scientific activities include computer control of industrial processes, model-based predictive control, robust control, and modelling and simulation of continuousprocesses. He has authored and co-authored more than 150 technical papers in interna-tional journals and conferences. Since 2002, he has been president of the SpanishAssociation of Automatic Control CEA-IFAC, where he promotes academic and industrialrelations.
